I will describe for formula the computer used to produce these teams.

We will be using the following data to form teams:

• Division the child played in last year
• Coach evaluations from last year
• Age
• Geography

Here is the actual calculations and the thought process that went into the formula.

The coaches rate the players on a scale from 1-10 which we converted to 1-100 when it was entered into the database. This was done by just multiplying the coaches numbers by 10. There were two reasons for this, first to give us more numbers to work with and to eliminate fractions which some coaches used. Since we accept evaluations from both head and assistant coaches we average these and treat each at equal weight. Children that did not play last year were given a default value of 60.

Coach 1	Coach 2	Coach 3	Average
30	60	40	43

The biggest variant we have is how one coach rates children compared to the other coaches. As I reviewed the results of the first tests of the balancing formula I compared children that I had coached in the past and found the results didn't map well to what I would have rated the two children relative to each other. The major difference was the coaches evaluations. We cannot eliminate this unless we can give the coaches a better structure for writing evaluations.

Another aspect of our evaluation methodology is we ask the coaches to rate the child based on how they compare to the other players in the division. This means that 7 year olds are being compared against 6 and 7 year olds and 8 year olds are being compared against 8 and 9 year olds. This is again going to skew the numbers slightly, ideally we would have wanted to compare these children with other 7 and 8 year olds.. To adjust for this we add 10 points to children that moved up a division this year. So in our example we assume our player has just been promoted to U10 and will therefore be given the penalty.

Coach Rating	First Year Penalty	Division Adjusted Rating
43	10	53

In the lower divisions the age of a player can have a pretty significant impact. In reviewing the coaches comments on their evaluation forms it is common to see something like this, "We had 7 first graders and only 2 second graders, everyone else had all second graders." In the upper divisions this probably has less impact.

To account for this we calculate the number of months old the child is normalized for the division. Players just qualifying to play in a division would have a month value of 0 whereas children that are just shy of moving to the next higher division would have month count of 24. Children that are playing up by parent request would have negative month counts. Here are some example children for U10 where the child should be born between Aug 1, 1990 and July 31, 1992.

Child 1 Sep 15, 1992	Child 2 July 15, 1992	Child 3 Aug 15, 1991	Child 4 Aug 15, 1990
-2 This child is playing up	0	12	24

We then take this number and subtract 12 from it and subtract this from our rating. So older children will lower their rating by as much as 12 points and younger children can raise theirs by 12 (or as many as 18 if playing up). In our example we will assume our player is on the young side and has a month count of 6, -6 once normalized.

Division Adjusted Rating	Age Adjustment	Final Rating
53	6	59

Now we have our ratings we just need to form teams in a geographic area. In the past we have relied exclusively on schools which resulted in our pairing children with coaches year after year. This violates one of the AYSO rules and regulations which states that the only team retention is the coaches children.

We have addressed this this year by adopting a strategy used by Sunnyvale Basketball. We have divided Sunnyvale into 4 quadrants. North and south of Fremont Ave and east and west of Sunnyvale-Saratoga/Matilda.

Zone 1 North East	Zone 2 North West	Zone 3 South East	Zone 4 South West	Zone 5 Else
Cherry Chase Challenger Cumberland Fremont Slater St Cyprian St Francis SV Middle SV Christian Vargas	Bishop Braly Briarwood Columbia Ellis Fairwood Lakewood Ponderosa San Miguel Rainbow Wilcox	Blach Cupertino Homestead Los Altos Montclaire Nimitz Oak South Penn Hebrew West Valley	Carden Collins Harker Laurelwood McAuliffe Peterson St Joseph Stocklmier Washington Open	Pre-schools, Out-of-town, Schools I could not find on the map.

The computer forms teams inside a zone pulling people based on their final rating starting with the lower numbers and working toward the higher it puts people on teams in a round table fashion. The players in the last zone meant to be manually put on teams in the other zones, treat them as a pool.